io#
This module contains methods to handle samna devices, IO and other convenience methods.
- calculate_neuron_address(x: int, y: int, c: int, feature_map_size: Tuple[int, int, int]) int[source]#
Calculate the neuron address on the devkit. This function is designed for ReadNeuronValue event to help the user check the neuron value of the SNN on the devkit.
- Parameters:
x (int) – x coordinate of the neuron
y (int) – y coordinate of the neuron
c (int) – channel index of the neuron
feature_map_size (Tuple[int, int, int]) – the size of the feature map [channel, height, width]
- Returns:
neuron_address (int)
- close_device(device_id: str)[source]#
Close a device by device identifier
- Parameters:
device_id (str) – device_name:device_id pair given as a string. dynapcnndevkit:0 or speck:0 or dynapcnndevkit:1
- disable_timestamps(device_id: str) None[source]#
Disable timestamps of the samna node
- Parameters:
device_id (str) – Name of the device to initialize. Required for different existing APIs for Dynapcnndevkit and Speck chips
- discover_device(device_id: str)[source]#
Discover a samna device by device_name:device_id pair
- Parameters:
device_id (str) – Device name/identifier (dynapcnndevkit:0 or speck:0 or dvxplorer:1 … ) The convention is similar to that of pytorch GPU identifier ie cuda:0 , cuda:1 etc.
- Returns:
device_info (samna.device.DeviceInfo)
- enable_timestamps(device_id: str) None[source]#
Disable timestamps of the samna node
- Parameters:
device_id (str) – Name of the device to initialize. Required for different existing APIs for Dynapcnndevkit and Speck chips
- events_to_raster(event_list: List, layer: int) Tensor[source]#
Convert an eventList read from samna to a tensor raster by filtering only the events specified by layer.
- Parameters:
event_list (List) – A list comprising of events from samna API
layer (int) – The index of layer for which the data needs to be converted
- Returns:
raster (torch.Tensor)
- events_to_xytp(event_list: List, layer: int) array[source]#
Convert an eventList read from samna to a numpy structured array of x, y, t, channel.
- Parameters:
event_list (List) – A list comprising of events from samna API
layer (int) – The index of layer for which the data needs to be converted
- Returns:
xytc (np.array) – A numpy structured array with columns x, y, t, channel.
- get_device_map() Dict[source]#
- Returns:
dict(str (samna.device.DeviceInfo))
Returns a dict of device name and device identifier.
- is_device_type(dev_info: DeviceInfo, dev_type: str) bool[source]#
Check if a DeviceInfo object is of a given device type dev_type
- Parameters:
dev_info (samna.device.DeviceInfo) – Device info object
dev_type (str) – Device type as a string
Returns –
-------- – bool
- launch_visualizer(receiver_endpoint: str, width_proportion: float = 0.6, height_proportion: float = 0.6, disjoint_process: bool = True)[source]#
Launch the samna visualizer in a separate process.
NOTE: MacOS users will want to use disjoint_process as True as a GUI process cannot be launched as a subprocess.
- Parameters:
receiver_endpoint (str) – the visualiser’s endpoint for receiving events (e.g. “tcp://0.0.0.0:33335”).
width_proportion (bool) – the rate between window width and workarea width of main monitor, default 0.75 which means this window has a width which equals to 3/4 width of main monitor’s workarea.
height_proportion (bool) – the rate between window height and workarea height of main monitor, default 0.75 which means this window has a height which equals to 3/4 height of main monitor’s workarea.
disjoint_process (bool, optional) – If true, will be launched in a disjoint shell process. Defaults to True. If false, this just runs the default samna command.
- Returns:
gui_process (Process) – The gui sub-process handle if disjoint_process was False.
- neuron_address_to_cxy(address: int, feature_map_size: Tuple[int, int, int]) Tuple[source]#
Calculate the c, x, y, coordinate of a neuron when the address of the NeuronValue event is given :param address: the neuron address of the NeuronValue event :type address: int :param feature_map_size: the size of the feature map [channel, height, width] :type feature_map_size: Tuple[int, int, int]
- Returns:
neuron_cxy (Tuple[int, int, int]) – the [channel, x, y] of the neuron